fp16ScalarProduct cudaMemcpy cudaFree cudaMallocHost cudaFreeHost cudaMalloc cudaGetDeviceProperties whole ./ ../ ../../../Common CUDA Runtime API FP16 Floating Point 16 true fp16ScalarProduct.cu FP16 1:CUDA Basic Topics sm60 sm61 sm70 sm72 sm75 sm80 sm86 sm87 sm89 sm90 arm sbsa x86_64 linux windows7 ppc64le linux x86_64 macosx 6.0 exe